Yesterday, after finishing a client proposal, my brain felt boiled over
At 10 p.m. Wednesday, I was revising a brand proposal in my Hangzhou rental. The AI’s copy kept getting more complete, but I could no longer pinpoint what was wrong: second draft, third draft, fourth draft, with a dozen windows open. By midnight, I stared at the screen and realized I wasn’t moving the work forward—I was just going in circles with it. I’d gotten stuck this way before: the more tools I added, the messier my brain became.
Now I do one thing: turn every conversation into a “task card”
This isn’t a new tool. It’s a plain method: define the goal and constraints first, then ask AI for options. Once I get the result, I review it against three questions: accurate, usable, and clear next step. If it isn’t right, I point out the specific gap instead of rewriting everything. My friend Lin Yue, a freelance photographer in Chengdu, puts “client shoot brief—reference images—delivery checklist” into one fixed card and copies it before starting each conversation. Even with a small client base, she avoids repeating avoidable work. Chen Chen, who leads a small website team in Shenzhen, keeps common task cards in a shared document so design and development start with the same context. The names and scenarios are fictional, but the method can be used immediately.
Replicating it today: ¥0, about 15 minutes, no coding needed
The cost is ¥0. Organizing the first card takes about 15 minutes; each task then takes two minutes to reuse. The technical threshold is simply being able to describe one thing clearly—no English or technical vocabulary required. First step: create one place where you regularly write prompts, such as Notes or a document. Title it “First Draft of Client Proposal,” then leave three lines below it: “What I want, what I don’t want, what I’ll do when it’s finished.” If you can’t write those three lines yet, don’t rush and don’t switch to pricier software. Writing them on a paper note works just as well.
Depending on where I’m busy, this is how I’d choose
If I’m just getting started, I’d make one task card with three sentences covering the goal, constraints, and deliverable. There’s no need to organize everything now; I’d wait until I get stuck twice on the same thing. If I already have one or two clients, I’d keep a fixed card for each, recording their tone, materials, and previous changes. Then each conversation starts by filling the gap—not infinite rewriting. If I’m scaling, I’d build a shared library of team templates and require clear acceptance criteria before anything goes to AI, so we get plenty of output but enough that is actually usable. For me, the real savings aren’t a few clicks. It’s the attention I can still spend working after AI has sent me in circles.